cuda_check_local_progress polled 1000000 times

From: Francesco Pietra (
Date: Sat Jun 02 2012 - 14:06:49 CDT

With namd-cuda 2.9 on a shared-mem machine with two GTX-580 (Debian
amd64) minimization (ts 0.1fs, wrap all) on a new system of a protein
in a water box, crashed at step 2,296 out of planned 10,000. Changing
to 2.9b3 multicore, the minimization worked well, ending at grad 1.5.
I did not notice if this known issue at the time of beta tests had
been fixed.

francesco pietra

If relevant at all; otherwise please skip the following:

I was at amber parm7, from LEaP prmtop/inpcrd, with some irregularities:

I used concanavalin-derived parameters for the Ca(II) ions present and
frcmod.ionsjc_tip3 for the monovalent and water. The protein was first
solvated (TIP3PBOX 12 0.85), then neutralized
(Na+ 0), and finally

addions x Na+ 200
addions x Cl- 200

Some 15 Cl- per box side were out of the box. Examination of the
periodicity (VMD) showed that also the outside Cl- were not mixed with
those of neighboring cell. On minimization, probably some mixing
occurred, and one Na+ and one WAT were also out, although there were
fewer total ions outside the box

Perhaps the worse is with charges: the final system (VMD) had charge
total -2.0596 (should have been 0.0000) and protein -38.99999. The
non solvated system had charge -23.99999 (which is correct). Before
adding ions, there was no WAT outside the box, so that one wonders how
Cl- ion got the outside. I posted to amber but it did not attract

The same system along the psf/pdb line showed no problem for the same
conc of salt (although with charmm I was unable to fix the ligand
partial charges the charmm-way with FFTK).


This archive was generated by hypermail 2.1.6 : Mon Dec 31 2012 - 23:21:36 CST